Agents

An Agent is the core building block in Fluo.
Agents are intelligent AI assistants you configure and then run inside your apps.

What makes up an Agent?

  • Persona & Output
    Define who the agent is and what kind of response it should return.

    • Persona: personality, tone, or style (friendly, concise, expert).
    • Output: decide the response format. You can define a custom schema so the agent produces structured results (JSON, summaries, tables, or plain text).

  • Purpose
    The role or goal of the agent (support, research, automation, analysis, etc.).

  • Instructions
    Additional rules or behavior guidance (e.g. “always answer in bullet points”).

  • Tools
    External integrations the agent can call — APIs, services, or apps (Slack, calendar, search, etc.).

  • Knowledge Bases
    Private documents and datasets the agent can reference.

  • Contexts
    Structured memory the agent can recall across conversations, sessions, or entities (like user profiles or project state).

Inputs

Agents aren't limited to just chat prompts. They can accept different forms of input depending on the problem you're solving. You can send text, upload files, provide images, or even stream audio into an agent. This flexibility lets you design agents for more than conversation — for analysis, automation, creative tasks, or data processing.

Flexible Inputs:

Agents in Fluo can take multiple input types:

  • Text (chat, instructions, prompts)
  • Audio (voice input)
  • Documents (PDFs, notes, transcripts)
  • Images (screenshots, diagrams, photos)

Outputs

When configuring an agent, you decide how its response should look. Outputs can be simple (like plain text) or structured (using a schema you define). This makes agents work not just as assistants but as predictable building blocks in larger systems — whether you want a short message, a detailed summary, or structured JSON you can feed into another workflow.

Custom Outputs:

You decide how your agent responds:

  • Plain text (chat style)
  • Structured data (custom JSON schema you define)
  • Rich outputs (summaries, tables, citations, etc.)

Not Just for Chat

Agents are not limited to chat interfaces.
You can run them as general-purpose AI functions inside your workflows, APIs, or apps. Whether the input is a block of text, an uploaded file, or an audio snippet, the agent processes it and produces the output format you define.

👉 Think of an Agent as your configurable AI function. You decide how it behaves, what it knows, and what it produces — Fluo handles the rest.